> ## Documentation Index
> Fetch the complete documentation index at: https://soran.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Asking questions

> Get an analyst-grade answer without building a query.

Soran AI sits at the top of your property's landing screen, and it is the first
thing there on purpose: most people arrive mid-task with a question rather than a
desire to browse.

Ask in plain English. "Why did Meta drop last week?" or "which products make the
most profit?" are the shape of question it is built for.

## What it does with a question

It reads the same reports you would have opened, through the same engine, so its
answers agree with the screens. It is not a separate estimate — if it tells you
revenue was a number, that is the number the attribution report shows for the
same period and model.

It can also take you somewhere. If the answer is best seen as a report, it can
navigate you to the right one with the right period and model already applied.

## Asking well

Name the period and the comparison you have in mind. "Compare last 30 days to the
30 before" is answerable directly; "how are we doing?" needs a follow-up before
anything useful comes back.

> [!TIP]
>
> If a number surprises you, ask it to show which report it came from. Reading
> the report yourself is how you check the answer, and the assistant is good at
> getting you to the right view of it.

## Related

> **[What Soran AI can see](/docs/soran-ai/what-it-can-see)**
>
> The data it can reach, and the data it cannot.

> **[Limits](/docs/soran-ai/limits)**
>
> Where it stops, and what to do instead.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.